NEW YORK – A team led by researchers at Stanford University has used Ionpath's spatial proteomics platform to identify characteristics of ductal carcinoma in situ lesions associated with the development of invasive breast cancer.
Detailed in a study published last week in Cell, the work provides insight into basic biology underlying DCIS and its links to invasive breast cancer and could ultimately point toward biomarkers that could help clinicians distinguish between high-risk and low-risk DCIS lesions, said Michael Angelo, senior author on the study and an assistant professor of pathology at Stanford as well as a founder of Ionpath.
DCIS is a pre-invasive form of breast cancer. DCIS lesions are not life-threatening, but as the Cell authors noted, as many as 50 percent of women diagnosed with DCIS will develop invasive breast cancer within 10 years, though it is not well understood whether this is due to the DCIS lesions themselves progressing to invasive cancer or some other mechanism linking DCIS to the future development of invasive cancer. In any case, because it is currently impossible to distinguish DCIS patients who will go on to develop invasive cancers from those who will not, nearly all women with DCIS are treated with surgery followed, in some cases, by additional treatments like radiation therapy.
DCIS has become a particularly pressing issue in breast cancer screening due to its high prevalence. Roughly 20 percent of new breast cancer diagnoses are DCIS. That figure has risen dramatically since the advent of widespread mammograms. Between 1983 and 2003, the incidence of DCIS increased 500 percent.
The Cell researchers set out to investigate whether profiling DCIS lesions using Ionpath's Multiplexed Ion Beam Imaging (MIBI) system would help them identify patterns that distinguished between high-risk and low-risk lesions. They analyzed a set of 58 primary DCIS tumors, 44 from patients who did not later develop invasive breast cancer and 14 from individuals who did progress to invasive breast cancer matched by age and year of DCIS diagnosis. Additionally, they analyzed nine normal breast tissue samples and 12 invasive breast cancer samples.
Menlo Park, California-based Ionpath's MIBIscope platform uses metal-conjugated antibodies and mass spectrometry to collect high-resolution spatial proteomics data. The platform is conceptually similar to Standard BioTools' (formerly Fluidigm) Hyperion imaging mass cytometry platform, but the MIBIscope uses an ion beam as opposed to a laser for ionizing samples and a secondary-ion mass spec with a time-of-flight analyzer for its sample analysis.
The platform allows for highly multiplexed protein measurements at the single-cell level. According to Brad Nelson, senior VP of marketing and corporate strategy at Ionpath, the MIBIscope can currently measure 40 proteins at a time. In the Cell study, the researchers looked at 37 proteins providing information on the cell types and cell function within the different tissues examined as well as their spatial organization. They generated a total of 433 parameters through this analysis that they used to distinguish between the DCIS lesions in patients who would go on to develop invasive breast cancer from those who would not.
The classifier they generated with that information distinguished between the two sets of patients with an area under the curve of 0.74. Interestingly, Nelson noted, spatial data played an outsize role in the classifier's predictive power, suggesting the utility of a spatial omics-based approach. The authors wrote that "spatially informed metrics describing cell densities, cell neighborhoods, pairwise cell distances, collagen structure, and multiplexed subcellular features" comprised 15 of the top 20 predictive features despite making up less than half of the features included.
Also notable, Angelo said, the most predictive features weren't based on the tumor cells themselves but rather the environment around the tumor cells — "What immune cells were there, what fibroblasts were around there," and, most significantly, the structure of what is known as myoepithelium, a layer of cells that sits between the DCIS lesion and the surrounding stromal tissue.
Somewhat counterintuitively, the researchers found that breakdown of this myoepithelial layer was associated with DCIS patients who did not go on to develop invasive breast cancer.
"Going into this, we were guessing that women who had the more contiguous, thick layer [of myoepithelium] would be the women who did well," Angelo said. "But of course, nothing is ever the way you think it will be. It was actually the exact opposite."
More generally, individuals who did develop invasive cancer had DCIS lesions that were more similar to normal tissue across a variety of measures.
Angelo said that more research is needed to make sure this result holds up in additional cohorts and, assuming it does, to explore what mechanisms might underly the phenomenon. He suggested, though, that it appears the more abnormal DCIS lesions could be stimulating an immune response that protects patients from developing invasive breast cancers in the future — a breast cancer vaccine of sorts. He noted that this theory is bolstered by the fact that the loss of myoepithelium that was in the Cell study also correlated with immune reactivity in the surrounding tissue.
"There were certain immune cells that seemed to go up, and there also seemed to be a stromal reaction in terms of fibroblast activity," he said. "Perhaps if you start to lose that myoepithelial barrier you start to unmask the tumor cells to the surrounding stroma so they can be more easily detected [by the immune system]."
Angelo highlighted a few limitations of the study, most notably the fact that it was done on retrospective samples and that all of the samples came from women who had undergone treatment for DCIS.
"These features are really telling us about women who have undergone treatment," he said. "That's really valuable information, but what we would really like to get at is to see if we see similar things in a prospective study where women either are or are not receiving surgical intervention, chemotherapy, or radiation. Because the most beneficial thing for patient care is going to be if we can find a predictive signature that will allow us to say at the time of diagnosis, 'This is a low-risk lesion; we're going to do watchful waiting,' versus, 'This is a high-risk lesion; we're going to do surgery and radiation and chemo.'"
There are several randomized controlled trials ongoing in the US and around the world looking at whether surveillance is a safe option for low-risk DCIS lesions including one, the COMET (Comparing Treatment Options for Women with Low-Risk Ductal Carcinoma In Situ) study being led by Shelly Hwang a surgical oncologist at Duke University and author on the Cell paper.
Angelo said that he and his colleagues hope to be able to participate in one of these studies. They are currently performing another retrospective analysis with roughly triple the samples of the Cell study in hopes of firming up their initial findings.
"We hope to have that study completed and to have preliminary data in six to eight months, and then in the meantime we are kind of getting our ducks in a row to do a prospective study," he said.